Specificity is an algorithm that calculates the weight that is applied to a given CSS declaration. The weight is determined by the number of selectors of each weight category in the selector matching the element (or pseudo-element). If there are two or more declarations providing different property values for the same element, the declaration value in the style block having the matching selector with the greatest algorithmic weight gets applied.

The specificity algorithm is basically a three-column value of three categories or weights - ID, CLASS, and TYPE - corresponding to the three types of selectors. The value represents the count of selector components in each weight category and is written as ID - CLASS - TYPE. The three columns are created by counting the number of selector components for each selector weight category in the selectors that match the element.

The selector weight categories are listed here in the order of decreasing specificity:

ID column

Includes only ID selectors, such as #example. For each ID in a matching selector, add 1-0-0 to the weight value.

CLASS column

Includes class selectors, such as .myClass, attribute selectors like [type="radio"] and [lang|="fr"], and pseudo-classes, such as :hover, :nth-of-type(3n), and :required. For each class, attribute selector, or pseudo-class in a matching selector, add 0-1-0 to the weight value.

TYPE column

Includes type selectors, such as p, h1, and td, and pseudo-elements like ::before, ::placeholder, and all other selectors with double-colon notation. For each type or pseudo-element in a matching selector, add 0-0-1 to the weight value.

No value

The universal selector (*) and the pseudo-class :where() and its parameters aren't counted when calculating the weight so their value is 0-0-0, but they do match elements. These selectors do not impact the specificity weight value.

Combinators, such as +, >, ~, " ", and ||, may make a selector more specific in what is selected but they don't add any value to the specificity weight.

The & nesting combinator doesn't add specificity weight, but nested rules do. In terms of specificity, and functionality, nesting is very similar to the :is() pseudo-class.

Like nesting, the :is(), :has(), and negation (:not()) pseudo-classes themselves add no weight. The parameters in these selectors, however, do. The specificity weight of each comes from the selector parameter in the list of selectors with the highest specificity. Similarly, with nested selectors, the specificity weight added by the nested selector component is the selector in the comma-separated list of nested selectors with the highest specificity.

The :not(), :is(), :has() and CSS nesting exceptions are discussed below.

Matching selector

The specificity weight comes from the matching selector. Take this CSS selector with three comma-separated selectors as an example:

[type="password"],
input:focus,
:root #myApp input:required {
  color: blue;
}

The [type="password"] selector in the above selector list, with a specificity weight of 0-1-0, applies the color: blue declaration to all password input types.

All inputs, no matter the type, when receiving focus, match the second selector in the list, input:focus, with a specificity weight of 0-1-1; this weight is made up of the :focus pseudo-class (0-1-0) and the input type (0-0-1). If the password input has focus, it will match input:focus, and the specificity weight for the color: blue style declaration will be 0-1-1. When that password doesn't have focus, the specificity weight remains at 0-1-0.

The specificity for a required input nested in an element with attribute id="myApp" is 1-2-1, based on one ID, two pseudo-classes, and one element type.

If the password input type with required is nested in an element with id="myApp" set, the specificity weight will be 1-2-1, based on one ID, two pseudo-classes, and one element type, whether or not it has focus. Why is the specificity weight 1-2-1 rather than 0-1-1 or 0-1-0 in this case? Because the specificity weight comes from the matching selector with the greatest specificity weight. The weight is determined by comparing the values in the three columns, from left to right.

[type="password"]             /* 0-1-0 */
input:focus                   /* 0-1-1 */
:root #myApp input:required   /* 1-2-1 */

Once the specificity values of the relevant selectors are determined, the number of selector components in each column are compared, from left to right.

#myElement {
  color: green; /* 1-0-0  - WINS!! */
}
.bodyClass .sectionClass .parentClass [id="myElement"] {
  color: yellow; /* 0-4-0 */
}

The first column is the value of the ID component, which is the number of IDs in each selector. The numbers in the ID columns of competing selectors are compared. The selector with the greater value in the ID column wins no matter what the values are in the other columns. In the above example, even though the yellow selector has more components in total, only the value of the first column matters.

If the number in the ID columns of competing selectors is the same, then the next column, CLASS, is compared, as shown below.

#myElement {
  color: yellow; /* 1-0-0 */
}
#myApp [id="myElement"] {
  color: green; /* 1-1-0  - WINS!! */
}

The CLASS column is the count of class names, attribute selectors, and pseudo-classes in the selector. When the ID column value is the same, the selector with the greater value in the CLASS column wins, no matter the value in the TYPE column. This is shown in the example below.

:root input {
  color: green; /* 0-1-1 - WINS because CLASS column is greater */
}
html body main input {
  color: yellow; /* 0-0-4 */
}

If the numbers in the CLASS and ID columns in competing selectors are the same, the TYPE column becomes relevant. The TYPE column is the number of element types and pseudo-elements in the selector. When the first two columns have the same value, the selector with the greater number in the TYPE column wins.

If the competing selectors have the same values in all the three columns, the proximity rule comes into play, wherein the last declared style gets precedence.

input.myClass {
  color: yellow; /* 0-1-1 */
}
:root input {
  color: green; /* 0-1-1 WINS because it comes later */
}

The matches-any pseudo-class :is(), the relational pseudo-class :has(), and the negation pseudo-class :not() are not considered as pseudo-classes in the specificity weight calculation. They themselves don't add any weight to the specificity equation. However, the selector parameters passed into the pseudo-class parenthesis are part of the specificity algorithm; the weight of the matches-any and negation pseudo-class in the specificity value calculation is the weight of the parameter's weight.

p {
  /* 0-0-1 */
}
:is(p) {
  /* 0-0-1 */
}

h2:nth-last-of-type(n + 2) {
  /* 0-1-1 */
}
h2:has(~ h2) {
  /* 0-0-2 */
}

div.outer p {
  /* 0-1-2 */
}
div:not(.inner) p {
  /* 0-1-2 */
}

Note that in the above CSS pairing, the specificity weight provided by the :is(), :has() and :not() pseudo-classes is the value of the selector parameter, not of the pseudo-class.

All three of these pseudo-classes accept complex selector lists, a list of comma-separated selectors, as a parameter. This feature can be used to increase a selector's specificity:

:is(p, #fakeId) {
  /* 1-0-0 */
}
h1:has(+ h2, > #fakeId) {
  /* 1-0-1 */
}
p:not(#fakeId) {
  /* 1-0-1 */
}
div:not(.inner, #fakeId) p {
  /* 1-0-2 */
}

In the above CSS code block, we have included #fakeId in the selectors. This #fakeId adds 1-0-0 to the specificity weight of each paragraph.

When creating complex selector lists with CSS nesting this behaves in exactly the same way as the :is() pseudo-class.

p,
#fakeId {
  span {
    /* 1-0-1 */
  }
}

In the above code block the complex selector p, #fakeId the specificity is taken from #fakeId and also the span, so this create a specificity of 1-0-1 for both p span and #fakeId span. This is the equivalent specificity as the :is(p, #fakeId) span selector.

Generally, you want to keep specificity down to a minimum, but if you need to increase an element's specificity for a particular reason, these three pseudo-classes can help.

a:not(#fakeId#fakeId#fakeID) {
  color: blue; /* 3-0-1 */
}

In this example, all links will be blue, unless overridden by a link declaration with 3 or more IDs, a color value matching an a includes the !important flag, or if the link has an inline style color declaration. If you use such a technique, add a comment to explain why the hack was needed.

Inline styles added to an element (e.g., style="font-weight: bold;") always overwrite any normal styles in author stylesheets, and therefore, can be thought of as having the highest specificity. Think of inline styles as having a specificity weight of 1-0-0-0.

The only way to override inline styles is by using !important.

Many JavaScript frameworks and libraries add inline styles. Using !important with a very targeted selector, such as an attribute selector using the inline style, is one way to override these inline styles.

<p style="color: purple">…</p>

p[style*="purple"] {
  color: rebeccapurple !important;
}

Make sure to include a comment with every inclusion of the important flag so code maintainers understand why a CSS anti-pattern was used.

CSS declarations marked as important override any other declarations within the same cascade layer and origin. Although technically, !important has nothing to do with specificity, it interacts directly with specificity and the cascade. It reverses the cascade order of stylesheets.

If declarations from the same origin and cascade layer conflict and one property value has the !important flag set, the important declaration is applied no matter the specificity. When conflicting declarations from the same origin and cascade layer with the !important flag are applied to the same element, the declaration with a greater specificity is applied.

Using !important to override specificity is considered a bad practice and should be avoided for this purpose. Understanding and effectively using specificity and the cascade can remove any need for the !important flag.

Instead of using !important to override foreign CSS (from external libraries, like Bootstrap or normalize.css), import the third-party scripts directly into cascade layers. If you must use !important in your CSS, comment your usage so future code maintainers know why the declaration was marked important and know not to override it. But definitely, don't use !important when writing plugins or frameworks that other developers will need to incorporate without being able to control.

The specificity-adjustment pseudo-class :where() always has its specificity replaced with zero, 0-0-0. It enables making CSS selectors very specific in what element is targeted without any increase to specificity.

In creating third-party CSS to be used by developers who don't have access to edit your CSS, it's considered a good practice to create CSS with the lowest possible specificity. For example, if your theme includes the following CSS:

:where(#defaultTheme) a {
  /* 0-0-1 */
  color: red;
}

Then the developer implementing the widget can easily override the link color using only type selectors.

footer a {
  /* 0-0-2 */
  color: blue;
}

Instead of using !important, consider using cascade layers and using low weight specificity throughout your CSS so that styles are easily overridden with slightly more specific rules. Using semantic HTML helps provide anchors from which to apply styling.

By indicating the section of the document you're styling before the element you're selecting, the rule becomes more specific. Depending on how you add it, you can add some, a lot, or no specificity, as shown below:

<main id="myContent">
  <h1>Text</h1>
</main>

#myContent h1 {
  color: green; /* 1-0-1 */
}
[id="myContent"] h1 {
  color: yellow; /* 0-1-1 */
}
:where(#myContent) h1 {
  color: blue; /* 0-0-1 */
}

No matter the order, the heading will be green because that rule is the most specific.

Reducing ID specificity

Specificity is based on the form of a selector. Including the id of an element as an attribute selector rather than an id selector is a good way to make an element more specific without adding an overabundance of specificity. In the previous example, the selector [id="myContent"] counts as an attribute selector for the purpose of determining the selector's specificity, even though it selects an ID.

You can also include the id or any part of a selector as a parameter in the :where() specificity-adjustment pseudo class if you need to make a selector more specific but don't want to add any specificity at all.

As a special case for increasing specificity, you can duplicate weights from the CLASS or ID columns. Duplicating id, class, pseudo-class or attribute selectors within a compound selector will increase specificity when overriding very specific selectors over which you have no control.

#myId#myId#myId span {
  /* 3-0-1 */
}
.myClass.myClass.myClass span {
  /* 0-3-1 */
}

Use this sparingly, if at all. If using selector duplication, always comment your CSS.

By using :is() and :not() (and also :has()), you can increase specificity even if you can't add an id to a parent element:

:not(#fakeID#fakeId#fakeID) span {
  /* 3-0-1 */
}
:is(#fakeID#fakeId#fakeID, span) {
  /* 3-0-0 */
}

Leveraging cascade layers is the standard way of enabling one set of styles to take precedence over another set of styles; cascade layers enable this without using specificity! Normal (not important) author styles imported into cascade layers have lower precedence than unlayered author styles.

If styles are coming from a stylesheet you can't edit or don't understand and you need to override styles, a strategy is to import the styles you don't control into a cascade layer. Styles in subsequently declared layers take precedence, with unlayered styles having precedence over all layered styles from the same origin.

When two selectors from different layers match the same element, origin and importance take precedence; the specificity of the selector in the losing stylesheet is irrelevant.

<style>
  @import TW.css layer();
  p,
  p * {
    font-size: 1rem;
  }
</style>

In the above example, all paragraph text, including the nested content, will be 1rem no matter how many class names the paragraphs have that match the TW stylesheet.

The best approach is to not use !important. The above explanations on specificity should be helpful in avoiding using the flag and removing it altogether when encountered.

To remove the perceived need for !important, you can do one of the following:

• Increase the specificity of the selector of the formerly !important declaration so that it is greater than other declarations
• Give it the same specificity and put it after the declaration it is meant to override
• Reduce the specificity of the selector you are trying to override.

All these methods are covered in preceding sections.

If you're unable to remove !important flags from an authors style sheet, the only solution to overriding the important styles is by using !important. Creating a cascade layer of important declaration overrides is an excellent solution. Two ways of doing this include:

Method 1

1. Create a separate, short style sheet containing only important declarations specifically overriding any important declarations you were unable to remove.
2. Import this stylesheet as the first import in your CSS using layer(), including the @import statement, before linking to other stylesheets. This is to ensure that the important overrides is imported as the first layer.

<style>
  @import importantOverrides.css layer();
</style>

Method 2

1. At the beginning of your stylesheet declarations, create a named cascade layer, like so:

   css

   @layer importantOverrides;
   

2. Each time you need to override an important declaration, declare it within the named layer. Only declare important rules within the layer.

   css

   [id="myElement"] p {
     /* normal styles here */
   }
   @layer importantOverrides {
     [id="myElement"] p {
       /* important style here */
     }
   }
   

The specificity of the selector of the important style within the layer can be low, as long as it matches the element you are trying to override. Normal layers should be declared outside the layer because layered styles have lower precedence than unlayered styles.

The proximity of an element to other elements that are referenced in a given selector has no impact on specificity.

body h1 {
  color: green;
}

html h1 {
  color: purple;
}

The <h1> elements will be purple because when declarations have the same specificity, the last declared selector has precedence.

Styles for a directly targeted element will always take precedence over inherited styles, regardless of the specificity of the inherited rule. Given the following CSS and HTML:

#parent {
  color: green;
}

h1 {
  color: purple;
}

<html lang="en">
  <body id="parent">
    <h1>Here is a title!</h1>
  </body>
</html>

The h1 will be purple because the h1 selector targets the element specifically, while the green is inherited from the #parent declarations.

In the following CSS, we have three selectors targeting <input> elements to set a color. For a given input, the specificity weight of the color declaration having precedence is the matching selector with the greatest weight:

#myElement input.myClass {
  color: red;
} /* 1-1-1 */
input[type="password"]:required {
  color: blue;
} /* 0-2-1 */
html body main input {
  color: green;
} /* 0-0-4 */

If the above selectors all target the same input, the input will be red, as the first declaration has the highest value in the ID column.

The last selector has four TYPE components. While it has the highest integer value, no matter how many elements and pseudo-elements are included, even if there were 150, TYPE components never have precedence over CLASS components. The column values are compared starting from left to right when column values are equal.

Had we converted the id selector in the example code above to an attribute selector, the first two selectors would have the same specificity, as shown below:

[id="myElement"] input.myClass {
  color: red;
} /* 0-2-1 */
input[type="password"]:required {
  color: blue;
} /* 0-2-1 */

When multiple declarations have equal specificity, the last declaration found in the CSS is applied to the element. If both selectors match the same <input>, the color will be blue.

A few things to remember about specificity:

1. Specificity only applies when the same element is targeted by multiple declarations in the same cascade layer or origin. Specificity only matters for declarations of the same importance and same origin and cascade layer. If matching selectors are in different origins, the cascade determines which declaration takes precedence.
2. When two selectors in the same cascade layer and origin have the same specificity, proximity is important; the last selector wins.
3. As per CSS rules, directly targeted elements will always take precedence over rules which an element inherits from its ancestor.
4. Proximity of elements in the document tree has no effect on the specificity.

• "Specificity" in "Cascade and inheritance"
• SpeciFISHity
• Specificity Calculator: An interactive website to test and understand your own CSS rules
• ID-CLASS-TYPE exercise a specificity quiz
• CSS key concepts:
  • CSS syntax
  • At-rules
  • Comments
  • Inheritance
  • Box model
  • Layout modes
  • Visual formatting models
  • Margin collapsing
  • Values
    • Initial values
    • Computed values
    • Used values
    • Actual values
  • Value definition syntax
  • Shorthand properties
  • Replaced elements
  • CSS nesting module